NVIDIA Accelerated IO (XLIO) Documentation Rev 3.60

Changes and New Features in this Version

Feature/Category

Description

Unified Configuration Subsystem (BETA)

Introduced a new JSON-based configuration system to simplify setup and improve the out-of-box experience - see XLIO Parameters.

Ultra API Server-Side Support

Added server-side support to the Ultra API for NVMe-TCP acceleration, enabling full functionality including listen and accept operations - see XLIO Ultra API.

Worker Threads Mode

Added a new Worker Threads Mode that enables XLIO to use internal worker threads for parallelized networking operations - see XLIO Execution Modes.

NGINX Dual-Port Support

Added support for dual-port configurations in NGINX deployments.

Obsolete API Removal

The following APIs have been removed:

  • SocketXtreme

  • Zero Copy recvfrom()

  • Packet Filtering Callback

Obsolete Features Removal

The following legacy features have been removed:

  • TAP support

  • NVME over TCP DIGEST Offload TX

  • SO_XLIO_ISOLATE option

  • CMSG_XLIO_IOCTL_USER_ALLOC option

  • Dummy Send

Important Notes

Note

As of XLIO v3.10, DEBUG-level logs will not be available in the release builds. Alternatively, you can use libxlio-debug.so to obtain those logs. For details on how to use libxlio-debug.so, please refer to Libxlio-debug.so section.

Note

libnl1 is no longer supported. Please use libnl3 instead.

Note

Direct Packet Control Plane (DPCP) provides a unified, flexible interface for programming NVIDIA NICs and comes as part of DOCA-Host. The DPCP version must be v1.1.44 and above. For further details, please see Direct Packet Control Plane (DPCP).

Note

TLS Rx offload related notes:

  • TLS Rx offload supports up to 64K concurrent connections

  • TLS Rx offload for IPv6 is not supported as it may cause unexpected behavior

  • TLS Rx offload requires OpenSSL 3.0.2 or higher and kTLS support from the kernel

© Copyright 2025, NVIDIA. Last updated on Nov 26, 2025